Introduction
 
Tipping is an interesting and special economic phenomenon which happens almost every day in most of countries. Since tipping is a completely voluntary thing, the reason why people leave the money for the server after the service is provided attracts lots of scholars to study. It challenges the point of view of standard neoclassical economic (Parrett, 2006) theory which assumed what humans will behave by making rational decisions. In the past studies and surveys, the economists demonstrated that people pay the tips in various reasons, normally because of social norm and feeling reciprocity and letdown aversion (Parrett, 2006) for the servers. Reciprocity refers to the idea that people reward kind actions and punish unkind actions, and implies a positive relationship between the size of the tip and service quality. The theory of letdown aversion asserts that decision-makers avoid letting others down (Parrett, 2006). For a server in the USA, the income paid by the firm is very low. One restaurant offers $2.25 per hours for the server. Thus, the large part of his/her income is from tips. As the investigation from U.S. Census Bureau showed that in 2003, sales at full-service restaurants were $151 billion. If we assum a tipping norm of 15%, the servers in the country could earn about $22.7 billion in tip income. Hence, it is important for a server to know how to get more tips. And there are plenty of determinants that influence the tipping size.
 
This paper explains why people leave money as tips after they are served and what are the determinants of the size of tips. I list a literature review of the research on tipping which provides old opinions and research results. And I also conduct a survey in a restaurant for a several days and get my data to see how gender of the customers, gender of the servers, bill size, group size, service quality, frequency, time, race and method of payment influence the tipping size.
 
Literature Review#p#分頁標題#e#
 
In theoretical research on tipping, the scientists mainly discussed about why people pay the tips after the service was given. The most common explanation for customer tipping is that it is a way to encourage the server to serve better and at the same time the customer can get better service, especially for people who frequently come by. Many research papers discussed this point such as Lynn (1996), Azar (2007). Ben-Zion and Karni (1977) were the economists to build the model of tipping. In their book, they also pointed out that people leave tips was associated with a repeated interaction between how much one tips and how better the next service will provide. And they also proved that customers who visit less frequently will tips less.
 
While their model couldn’t explain the motivation of customers to tip who were infrequently patronized. So some economists use utility function to link the servers’ incomes to consumers’ utility. They demonstrate that people are altruistically inclined, so there is a tendency for wanting to help those in need. It is called altruistic behavior that is one of the variables that will maximize consumers’ alternative utility function. Waiters and waitresses are often people who earn lower income, so tipping can be a form of altruism. Customers can get utility from helping others to make up the lost utility from giving extra money for tips. Hausman and Mcpherson(1996) built up a utility function that includes altruism can be specified as U = f (X, Y, Z, A…), where X, Y and Z are material goods and A is altruism.
 
There also have some explanation of social norm or social approval (Azar 2004b; Ruffle, 1999). To explain why people tip, it is said that people will tip because of social norm and get social approval. Disobeying social norms will result in a psychological disutility. This psychological disutility mainly from the other people come with you who will give social pressure on tippers. Or it main from the waiter especially you come alone. In Azar’s article, he established utility function which is related both traditional components and social norm component. The utility function is given by u(g; , θ) = d(g – ) + θp(g) – bg. “The norm in period t is denoted by , where represents the appropriate tip( as a percentage of the bill) and g is the tip in percentage of the bill, d is a function representing the disutility from social disapproval, b is the bill size, and p is the utility from tipping that arises from feeling generous, impressing others, and so on.” So the customer will maximizes utility by choosing g.
 
In empirical research on tipping, economists usually randomly pick up 100 to 200 data either on survey of servers or customers in restaurants (eg. Lynn and Grassman,1990), while the research done by Lynn and McCall (2000), which combines seven published and six unpublished studies which contain 2,547 dining parties at 20 different restaurants. So the results may more convincing. The mainly variables are bill size, tip size, service quality, gender, group size and so on.#p#分頁標題#e#
 
Gender of the customers and servers
 
According to Lynn(2006), Men sometimes leave larger tips than do women (e.g., Crusco and Wetzel, 1984; Lynn and Latane, 1984) and waitresses sometimes receive larger tips than do waiters (e.g., Davis, 1998). It is because men gain more income than women as a lot of studies showed. But Lynn and Graves, 1996; Lynn and Simons, 2000 didn’t find these sex effects on tipping. It also appears that the effect of customer sex on tipping depends on server sex and vice versa. For example, Conlin, Lynn and O’Donohue (2003) found a significant interaction between server and customer gender such that women tipped more than men when the server was male but not when the server was female. The main conclusion is that tipping is affected by the dynamics of sexual attraction.
 
Bill Size
 
There is a tradition in tipping size in the United States that the tips is normally around 15% to 20% of the bill, so the tip size are probably positively related to bill size. Lynn and McCall (2000) found that 69 percent of the average variability in tips can be explained by bill size alone. It means that bill size can be the most significant factor that determines the tip size.
 
Group Size
 
Most research found that the bigger group size usually leave less percentage of tips than small group size. For example, in Parrett(2006)’s research, he examed the relationship between tipping size and table size. And found that tippers may tip a higher amount in the presence of others at the table in order to gain social status. People are always not willing to lose face in front of others. But there also have tippers may "free ride" on the tips of others at the table, and thus tip less. It means when the waiter only sees the total tip, and not its division among the customers, somebody will reduce their tips in order to free-ride on others in the table. Giving fewer tips may not be as obvious or as embarrassing as compared to when there are fewer people. “However, the common restaurant practice of adding an automatic service charge onto bills at large table sizes (usually six or more customers) suggests that incentives to free ride may dominate”.(Parrett, 2006)
 
Service Quality
 
Service quality can be a very important variable. People tip mainly because of getting better service. Served by a better quality, customers will willing to give more tips. Lynn and McCall(2000) found that there is a statistically significant and positive relationship between service evaluations and tip size. But the effect of service on tips was small, however, accounting for less than 2% of the variability in tip percentage. They also discovered that tipping was not significantly related to servers' or third parties' evaluations of the service. This may cause servers to think that tips are not related to the service quality they provide, eliminating their incentive to make effort and resulting in an inefficiency of tipping as an instrument to improve service quality. Another research done by Bodvarsson and Gibson (1997) found positive correlation between service quality and tip size in five out of seven restaurants. Among these five restaurants, the correlation was statistically significant in three restaurants. The other two have negative correlation.#p#分頁標題#e#
 
Frequency
 
Lynn and McCall (2000) suggested that the frequently customers tend to base their tips more on bill size, probably because they know more about the percentage of paying the tips in the restaurant. And they will also leave more tips than infrequent patrons, because they are more likely to identify with servers or because they value servers’ approval more than do infrequent patrons (Lynn, 2006).
 
Race
 
Black restaurant patrons are more likely than White patrons to tip a flat amount rather than a percentage of the bill. Blacks also leave smaller average restaurant tip percentages than do Whites. This latter effect remains sizable and statistically significant after controlling for education, income and perceptions of service quality, so Black-White differences in tipping are not due solely to socio-economic differences or to discrimination in service delivery (Lynn 2006). In Lynn(2003), she found that Whites know more about the percentage customer usually pay for tips in United States than Black do, which was 71% versus 37%. Other research also showed that blacks and Asians are poor tippers. One possible explanation is that black or other ethnic minorities are less familiar with the tipping norm. Another possible explanation is the income effect on tipping Ethnic minorities tend to be in lower income groups. They also have larger family sizes and more dependent children in general. A third explanation is racial discrimination. If servers believed that ethnic minorities are poor tippers, they are not likely to provide higher quality service to those customers (FungFong,2005).
 
Method of payment
 
In Feinberg (1986)’s research, he found that people who pay with credit cards generally give larger percentage tips than those who pay with cash. And he listed several reasons that made this: (1) the delayed payments can somewhat reduce psychological cost. (2) “pre-existing differences between cash and credit-card customers. (3) conditioned responses to credit-card stimuli” (Feinberg, 1986).
 
Food Quality
 
Lynn and McCall (2000) found significant positive correlation between tip size and food ratings. But Lynn and Grassman (1990) found no such correlation. Additionally, some author found the positive correlation can only find from the survey given to the servers but not according to customers’ responses.
 
Empirical model
 
I present a production function as the empirical model. Assume that the tipping size will depend on following variables: Server’s Gender (SG), Women’s percentage in a table (WP), Bill size (BS), Time, Group size (GS), Service quality (SQ), Race, frequency (Freq), method of payment (MOP), Food quality (FQ), Dress, Server’s self introduce (Intro), and SG*WP (sqwp). So the model can be derived into:
 
T= f (SG, WP, BS, time, GS, SQ, Race, Freq, MOP, FQ, Dress, Intro, sqwp)#p#分頁標題#e#
 
Based on prior research, I believe that female servicer can receive more tips than male. And male customers usually pay more tips than female. Female server can receive more tips from male customers than from female customers. Also, the bigger the bill size, the more tips will be. Additionally, the server gets more tips at diner time than at lunch time. And I expect the bigger group size, the fewer tips as a percentage of total bill the server will receive. I also expect that a better service quality will gain more tips. And white people pay more tips than other races. The frequently visited customers usually pay more tips than people who come less frequently. Lastly, I assume that people pay more tips when they use credit card and people will pay more tips when the food quality is better.
 
All of the exogenous variables and their coefficients will be tested by multiple regression models to exam relationships between tipping size and other factors. T-Test will be a good method to see the results are significant or not. During the regression test, it will prove how every factor influence the tipping size and remove the unrelated factors.
 
Data
 
In order to know what factor will lead the size of tips, we should exam the relationship between possible variables and tipping size. All my data was collected in a restaurant in Ogden, Utah named Fuji Dragon. This restaurant is served with Chinese and Japanese food which attracts people from different countries. Servers there helped me to fill out my survey and recorded their working performances and their customers’ feedback. The total number of observations is 52. I have got right now. There are 12 questions appeared on my survey which contains how many people were at the table, how much the total bill was, do you introduce yourself to the table and so on. After collecting all the data, I processed the data and set variables as following:
 
Gender: Male=0, Female=1
 
Time: Lunch=0, Dinner=1
 
Frequency: 1-2 times=0, several=1, often=2
 
Method of payment: Cash=0, Credit Card=1
 
Food Quality: Excellent=3, good=2, ok=1, not good=0
 
Service quality: 1(poor) to 10(excellent)
 
Dressing: Very expensive=3, Somewhat expensive=2 , Average=1, Less Expensive=0
 
Race: White customers=1, Other races=0
 
Self introduce: yes=0, No=1
 
Results
 
I put all my variables into one regression, setting tip size as dependent variable. And I get the results as follows.
 
Regression Analysis: Dependent Variable = Tip Size
 
The regression equation is
 
T = 0.34 - 2.33 SG - 0.0199 WP + 0.133 BS + 0.540 Freq + 0.141 MOP + 0.862 FQ
 
- 0.124 SQ + 0.302 Dress - 0.408 Intro + 0.0229 sgwp + 0.829 white
 
+ 0.617 time - 0.105 GS#p#分頁標題#e#
 
Table-1
 
PredictorCoefSE CoefTP
 
Constant0.3392.3070.150.884
 
SG-2.3311.481-1.570.123
 
WP-0.019860.02199-0.900.372
 
BS0.133100.023595.640.000
 
Freq0.54000.37271.450.155
 
MOP0.14070.75480.190.853
 
FQ0.86190.47561.810.078
 
SQ-0.12430.2514-0.490.624
 
Dress0.30180.44430.680.501
 
Intro-0.40840.6418-0.640.528
 
sgwp0.022940.023830.960.342
 
white0.82880.76271.090.284
 
time0.61700.66610.930.360
 
GS-0.10510.2390-0.440.663
 
S = 1.55919 R-Sq = 75.0% R-Sq(adj) = 66.7%
 
Analysis of Variance
 
Source DF SS MS F P
 
Regression 13 284.913 21.916 9.02 0.000
 
Residual Error 39 94.812 2.431
 
Total 52 379.725
 
The regression result shows that the P-value of server’s gender, bill size, frequency and food quality are very significant. While other variables are not that significant or have no influence on tipping size.
 
The coefficient of server’ gender is negative 2.331. Since I set male to 0 and female to 1, it means that male servers can earn more tips than female servers when other variables keep constant. The coefficient of women’s percentage in a table is negative as well. It means that higher rate of women percentage in a table will decrease tipping size. It also means male customers usually pay more tips than female, but the influence is not too significant. The coefficient of the product of server’s gender and women percentage is positive. It explains that people pay more tips to the servers who have the same gender as themselves. However, this result doesn’t match my hypothesis. The main reason can be that most of the servers in my survey are female and the P-value of sgwp is 0.342 which is not very significant as well.
 
The coefficient of bill size is 0.13310 in this regression and the P-value is 0.000 which is very significant. It means that bill size will significantly influence the tipping size and they have a positive relation. The coefficient of frequency is positive too. It demonstrates that the frequently visited customers will pay more tips than in frequent customers. The coefficient of method of payment is 0.1407 which has positive relation between method of payment and tip size. It means that people who pay with credit card pay more tips than people who pay in cash.
 
The coefficient of food quality is 0.8619 and its P-value is as high as 0.078. It presents that the good food quality will bring a higher tips. While my result of service quality is negative 0.1243 in coefficient. It means that the good service quality won’t help the servers to gain more tips. The result is different from literature review which showed a positive correlation between service quality and tip size. But the P-value and T-test of service quality are not significant. And the service quality is rated by the servers own. It may have the gap between servers themselves feelings and the customers’ feelings.#p#分頁標題#e#
 
The customers’ dressing also has positive influence on tipping size. And server’s self introduce can also help increase the tip size. But both of them are not significant.
 
The coefficient of the race is positive which presents that the while customers do tip more than other races. But the P-value is not significant enough. The coefficient of time is 0.6170 which shows that at dinner time, the server can gain more tips than at lunch time. The coefficient of group size is -0.1051, it means that the bigger the group size the smaller the tips will be. It matches parrett’s result that people will free ride on others when in a big group.
 
According to this regression, I pull out the 4 most significant and focused variables to do a second regression to see how these four important variables influence the tip size.
 
Regression Analysis: T versus SG, BS, Freq, FQ
 
The regression equation is
 
T = - 0.878 - 1.08 SG + 0.145 BS + 0.443 Freq + 0.784 FQ
 
Table-2
 
PredictorCoefSECoefTP
 
Constant-0.87790.8574-1.020.311
 
SG-1.08270.4697-2.310.026
 
BS0.144500.015109.570.000
 
Freq0.44260.25861.710.093
 
FQ0.78420.34852.250.029
 
S = 1.49249 R-Sq = 71.8% R-Sq(adj) = 69.5%
 
Analysis of Variance
 
Source DF SS MS F P
 
Regression 4 272.804 68.201 30.62 0.000
 
Residual Error 48 106.922 2.228
 
Total 52 379.725
 
From the second regression we can see that all the four variables are statistically significant. The P-values are all less than 0.1. It proved that servers’ gender, bill size, frequency and food quality are all strongly influence the tip size.
 
The P-value of server’s gender is 0.026 and the absolute value of T-test is 2.31. The coefficient of server’s gender was negative 1.0827. So it means that the server’s gender is negatively related to tipping size. In other word, customers will give more tips to male servers than female servers. And male server can receive about $1 more than female server when other variables no changes.
 
The P-value of bill size is still 0.000, and T-test is 9.57. The coefficient of bill size in this regression is 0.14450. This means that when bill size increases by $1, the tips will increase by 14.45 cents if keep other variables unchanged. It proves that people do tip as the tipping norm at about 15%.
 
The P-value of frequency is 0.093 and the T-test is 1.71. The coefficient of frequency is 0.4426. The return customers do know the server well and they want to get good service next time, so they may probably give more tips than others do. At the same time, the server may also pay more attention to those customers and offer better quality. So the frequent customers usually pay more tips at a high rate.#p#分頁標題#e#
 
The P-value of food quality is 0.029 and the T-test is 2.25. The coefficient of food quality is 0.7842. It proves that the customers do add their feelings about the food to the tip size. When they like the food, they are more willing to pay the tips. On the contrary, they will reduce the tips if they feel the food quality was not so satisfied. While this factor can not controlled by the servers.
 
Conclusion
 
Tipping behavior is an interesting topic and special economic phenomenon. To find the reasons and the determinants that influence the size of tips is a very good experience and worth to study with. After a series of study with this topic, I get my own result. To answer do people give tips based on a social norm or as a reward for good service, can be concluded as follows.
 
First of all, tip size is not related to service quality. Although people tip to expect good service quality, as the research found that the service quality will not determine the tipping size.
 
Secondly, male servers, bill size, and frequency of visit are positively related to tip size. Thus, social norms may dictate tipping behavior. As the result shows, people generally tip around 15% of the bill size which matches the tipping norm in the United States. It means people pay more attention on social norm than any other experience. Additionally, the male servers and the frequency of visit can also help to gain additional tips.
 
Thirdly, because food quality is an important component of tipping behavior, the total customer experience rather than service quality is important. The result demonstrates that food quality is more important than service quality. When people decide to pay tips, they will not only judge servers’ service quality, but also consider the food quality. It also proves that people will think over the total experience not only consider one or two aspects.
 
Appendix A: Copy of Survey
 
1. How many people were at the table?
 
2. How many people were men and how many were women?
 
Men____ Women___
 
3. How much was the total bill?
 
4. Were they the frequency customers?
 
(1-2 time____ several times____ frequently____)
 
5. Was it at lunch or dinner? (lunch__ dinner___)
 
6. Was everything perfect while you were serving?
 
7. What race your customers are?
 
(White____ Asian___ black or African American___ Hispanic or Latino ethnicity)
 
8. How they paid? (Credit card or cash)
 
9. Did you ask if the food was OK? If so how did they respond
 
(Excellent___ Good____ OK_____ Not Good____)
 
10. Did you introduce yourself to the table?
 
11. Please rate how expensive the clothing was of the people at the table#p#分頁標題#e#
 
(Very expensive____ Somewhat expensive____ Average___ Less Expensive____)
 
12. On a scale of 1(poor) to 10(excellent) evaluate the service you gave to the table.
 
References
 
Parrett, Matt. (2006, October 1). An analysis of the determinants of tipping behavior: a laboratory experiment and evidence from restaurant tipping The Free Library. (2006). Retrieved January 25, 2012 from http://www.thefreelibrary.com/An analysis of the determinants of tipping behavior: a laboratory...-a0154513788
 
Ben-Zion, Uri and Edi Karni (1977), "Tip Payments and the Quality of Service," in O.C. Ashenfelter & W.E. Oates,(Eds.), Essays in Labor Market Analysis, New York: John Wiley & Sons, pp. 37- 44.
 
Azar, O.H. 2004a. “The History of Tipping – from Sixteenth-Century England to United States in the 1910s.” Journal of Socio-Economics 33: 745-764.
 
Azar, O.H. 2004b. "What Sustains Social Norms and How They Evolve? The Case of Tipping." Journal of Economic Behavior and Organization 54: 49-64.
 
Azar, O. H. (2007), The Social Norm of Tipping: A Review. Journal of Applied Social Psychology, 37: 380–402. doi: 10.1111/j.0021-9029.2007.00165.x
 
Lynn, Michael (2006), “Tipping in restaurants and Around the Globe: An Interdisciplinary Review.” Ch. 31, pp. 626-643. In Morris Altman (Ed.) Handbook of Contemporary Behavioral Economics: Foundations and Developments, M.E. Sharpe Publishers.
 
Feinberg, Richard A. (1986), “Credit Cards as Spending Facilitating Stimuli: A Conditioning Interpretation,” Journal of Consumer Research, 13 (December), 348-356.
 
Ruffle, Bradley J. (1999), "Gift Giving with Emotions," Journal of Economic Behavior and Organization, 39, 399-420.
 
Lynn, Michael and Michael McCall. 2000. "Gratitude and Gratuity: A Meta-Analysis of Research on the Service-Tipping Relationship." Journal of Socio-Economics 29: 203-214.
 
Lynn, Michael (1996), “Seven Ways to Increase Server’s Tips,” Cornell H.R.A. Quarterly, (June), 24-29.
 
Lynn, Michael (2003), “Restaurant Tips and Service Quality: A Weak Relationship or Just Weak Measurement?,” International Journal of Hospitality Management, 22, 321-325.
 
Conlin, Michael, Michael Lynn and Ted O'Donahue. 2003. "The Norm of Restaurant Tipping." Journal of Economic Behavior and Organization 52: 297-321.
 
Lynn, Michael and Andrea Grassman. 1990. “Restaurant Tipping: An Examination of Three 'Rational' Explanations,” Journal of Economics Psychology 11:2, pp. 169-81.
 
Bodvarsson, Orn B. and William A. Gibson. 1997. “Economics and Restaurant Gratuities: Determining Tip Rates,” American Journal of Economics and Sociology 56:2, pp. 187-204.#p#分頁標題#e#